Schedule disruptions represent a major challenge for commercial aviation. The complexity of airline operations coupled with tight operational margins and numerous constraints, imposed by route-network and regulatory limitations, make disruption recovery especially challenging. Disruption recovery research is aided by datasets, allowing algorithms and systems to be tested on real-world data. We present a large-scale dataset, featuring multiple hubs, combining actual flight data from the US route-network and simulated passenger itineraries. We evaluate its properties and test two algorithms, comparing their results on this dataset to the previous European based, single hub, small-scale dataset.
|State||Published - 31 Dec 2016|